$ONCONFIG Parameter: RTO_SERVER_RESTART

IDS 11.50.FC5 AIX 5.3 While reviewing the $ONCONFIG parameters available on this instance, I came across this new parameter: ## Recovery Time Objective Server Restart RTO_SERVER_RESTART 0 # Number of Seconds, Overrides Checkpoint Interval ## Shared Memory Space Parameter CKPTINTVL 60 # Check point interval (in sec) Has anyone played with this parameter yet? Any downside to setting this parameter? Initial setting: same as the "now-defunct (overriden)" CKPTINTVL? Thanks in advance. Clifton _________________________________________________________________ Bing brings you maps, menus, and reviews organized in one place. http://www.bing.com/search?q=restaurants&form=MFESRP&publ=WLHMTAG&crea=TEXT_MFES RP_Local_MapsMenu_Resturants_1x1

RTO_SERVER_RESTART sets the recovery time goal that the engine uses to autonomically adjust the CKPTINTVL and LRU_MIN/MAX_DIRTY settings. It informs the engine that in the event of a hard crash you want the instance to be able to recover in RTO_SERVER_RESTART seconds or less. The shorter this value the more aggressively the engine has to checkpoint and flush data during and between checkpoints. I have a couple of clients that have set this without adverse effect on production performance, however, I would warn against making it too aggressive. Setting it to a 10 second recovery time might be bad. Art Art S. Kagel Oninit (www.oninit.com) IIUG Board of Directors (art@iiug.org) Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on my employer, Oninit, the IIUG, nor any other organization with which I am associated either explicitly or implicitly. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves. On Mon, Nov 2, 2009 at 2:36 PM, Clifton Bean <clifton_bean@hotmail.com>wrote: > IDS 11.50.FC5 AIX 5.3 > > While reviewing the $ONCONFIG parameters available on this instance, I came > across this new parameter: > > ## Recovery Time Objective Server Restart > RTO_SERVER_RESTART 0 # Number of Seconds, Overrides Checkpoint Interval> > ## Shared Memory Space Parameter > CKPTINTVL 60 # Check point interval (in sec)> > Has anyone played with this parameter yet? Any downside to setting this > parameter? > > Initial setting: same as the "now-defunct (overriden)" CKPTINTVL? > > Thanks in advance. > > Clifton > > _________________________________________________________________ > Bing brings you maps, menus, and reviews organized in one place. > > > http://www.bing.com/search?q=restaurants&form=MFESRP&publ=WLHMTAG&crea=TEXT_MFES RP_Local_MapsMenu_Resturants_1x1 > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --0023545bd64cb960880477696842

I set this parameter when we migrated to 11.50 and am very happy so far. RTO_SERVER_RESTART is the number of seconds you want fast recovery to take. IDS will make an educated guess as to how long fast recovery will take by keeping track of things like how fast you can write to your logical and physical logs (kept persistent via RAS_PLOG_SPEED and RAS_LLOG_SPEED in the ONCONFIG) and how much data you are writing to these logs. When the amount of time IDS estimates fast recovery will take reaches RTO_SERVER_RESTART it will fire a checkpoint. You can monitor the currently estimated time to perform fast recovery via onstat -g ckp </home/informix> onstat -g ckp IBM Informix Dynamic Server Version 11.50.FC3 -- On-Line (Prim) -- Up 68 days 02:38:53 -- 3740228 Kbytes AUTO_CKPTS=On RTO_SERVER_RESTART=90 seconds Estimated recovery time 43 seconds The idea is that you only issue a checkpoint when you need to in order to maintain a recovery time objective (RTO) instead of always every X secconds. The result is fewer checkpoints during low activity and more checkpoints during high activity and a way to "guarantee" the amount of time it will take to recover from a crash. You can turn this feature off (RTO_SERVER_RESTART 0) or set it to a number of seconds between 1 minute and 30 minutes. If you need to recover quickly from a crash, set it on the lower end (I'm currently set at 90 seconds) and you will have more checkpoints. If you don't care if it takes a while to recover from a crash, set it on the higher end and you will have fewer checkpoints (and probably better throughput). It is also a good idea to enable AUTO_LRU_TUNING with this feature. If during a checkpoint the engine realizes it would not be able to meet the RTO time in the event of a crash it will tune your LRUMIN/LRUMAX to flush the LRU queues to disk more agressively in between checkpoints. Hope this helps, Andrew